I really appreciate teachers like you with which we can understand the concepts and apply our research questions easily. I would request a video in detailing the difference between FE, RE, and ME. And also when to use these three models. Thanks you for your time.
I greatly appreciate it!! I have taken longitudinal courses but some of the concepts did not make sense to me. Now thanks to you I have a better vission of mixed effect models. (I am very grateful of you for teaching concepts with graphs)
Excellent video! Thank you. Would you kindly clarify why you used the measure at time 0 as one of the repeated measures? Shouldn't the value at time 0 (pre-treatment) be considered a covariate and not one of the repeated measures (i.e., post-treatment outcome)?
Thank you for such a great explanation! Two questions: 1. Should weeks be continuous or a factor variable? I guess it depends on if you want overall effect, or effect from week to week? When doing factor, I get an error when including weeks as random slope, how can I fix this? 2. How about if we want to see if the change is different depending on gender? Could we write: weight ~ weeks*gender + (weeks | subjects)? I am confused since the gender is essentially included in subject, but I want to see if there is any difference in change over time
Am I right if I say that including WeeksxDiet allows for DietA and DietB to have different slopes i.e.Subject 1 and Subject2 will have the same slope but it will be different from the slope of Subject3 and Subject4 (which have same slope). Further, in Weight ~ Weeks × Diet + (1 + Weeks|Subjects), including Weeks as a random effect will then allow all subjects to have uniques slopes which can be seen as a step further than just including WeeksxDiet but no random effect for Weeks?
thank for great video. I have a question that my data is repeated measurment, but the output values are not linear with the predictor, which method should I use?
If your problem fits the example in this video, except that there seems to be some nonlinear trend in the data, then you can use a nonlinear mixed effect model. I aim to create such a video next year.
Thanks for the great video! I'm trying to code along in R, but I can't replicate the simple regression. Is your code available anywhere? Or could you show me the data you used to fit the simple regression? That would be very helpful!
In the example, I have weeks and diet. Do you mean what would happen if you instead have two continuous variables? If so, you need a three dimensional plot, where you instead fit a plane to the data.
@@tilestats Thank you. yes I am asking about two continuous variables. I'm aware that they would form a plane but could this still be visualized in a similar way as above in two dimensions?
I really appreciate your clear explanation on the concept of linear mixed effects models. Could you or someone clarify which data format should be set for using lmer? I think 'long' format according to "weeks" would be appropriate. Right?
6:57 I don't understand how we can now assume that the Groups have one random intercept. That would mean each group has their own distribution, which is not the case because as you said all 4 subjects are randomly sampled out of one distribution.
When we include Diet in the model, we can test if the two groups have different intercepts. Thus, we then no longer assume that all individuals are sampled from the same distribution.
If I have different Genes count in place of weight How can I used that for 10-20 types Genes. Should I use a loop to get results for each Genes count changes over the week by the influence of Diet or is that any possibility to create model for all Genes at a time. One more question you have not explained how we will know how much changes happen in weight over the weeks while comparing both the diet
I would suggest that you look into how to set up a repeated measure design in packages, like DESeq2. limma or edeR, that have been developed to deal with gene count data.
Many thanks for your excellent videos. But the lecture over it is not much understandable. So, as feedback, can you possibly give your lecture's script to a text-to-speech engine (bot) to read it aloud over the video? I am saying this because you pronounce many words very badly, severely ruining your awesome video and badly distracting the audience. The worst of them is the word "model" which is pronounced by you as "mole" or "mowel". There are many such odd pronounciations. I would be very grateful if you could give the text to a bot to speak it. Thanks again for your awesome videoes.
What a cringy comment dude. Thats his pronunciation, which is fine and understandable, what do you mean "you get distracted" by it? Pay your respect to the man and stop distracting us with your useless and offensive comments.
This is the best video so far I have seen. We need people who can explain concepts in simple terms.
Excellent teacher! Thank you for all your efforts to make us understand it well! Well done!
I really appreciate teachers like you with which we can understand the concepts and apply our research questions easily. I would request a video in detailing the difference between FE, RE, and ME. And also when to use these three models. Thanks you for your time.
You are an excellent instructor!
Thank you very much for your content!!
Best explanation on mixed-effects models on RUclips. Thank you for a great job!
I greatly appreciate it!! I have taken longitudinal courses but some of the concepts did not make sense to me. Now thanks to you I have a better vission of mixed effect models. (I am very grateful of you for teaching concepts with graphs)
Thank you so much. I appreciate your clear explanation and example code. Very helpful.
Great explanation, thanks!
What an expert you are ! If you add some examples with animated figure, then you would be a super star to the academic field, I am sure.
adding videos is very time consuming, I think that's better to continue adding more videos covering a wider range of topics😉
Excellent video! Thank you.
Would you kindly clarify why you used the measure at time 0 as one of the repeated measures? Shouldn't the value at time 0 (pre-treatment) be considered a covariate and not one of the repeated measures (i.e., post-treatment outcome)?
Excellent video. Do you know how to get SPSS to produce the individual slopes as a new parameter?
Great video! By any chance do you have the R translated to Python codes?
Your videos are addictive! If you could do something about survival analysis that would be amazing, as there isn't much available yet on RUclips.
ruclips.net/p/PLLTSM0eKjC2cj7XtMM6OuX9zsvBava1FI
Thank you for such a great explanation!
Two questions:
1. Should weeks be continuous or a factor variable? I guess it depends on if you want overall effect, or effect from week to week? When doing factor, I get an error when including weeks as random slope, how can I fix this?
2. How about if we want to see if the change is different depending on gender? Could we write: weight ~ weeks*gender + (weeks | subjects)? I am confused since the gender is essentially included in subject, but I want to see if there is any difference in change over time
Weeks should be continuous. You can include gender just as I include Diet A/B in the example in this video.
Am I right if I say that including WeeksxDiet allows for DietA and DietB to have different slopes i.e.Subject 1 and Subject2 will have the same slope but it will be different from the slope of Subject3 and Subject4 (which have same slope).
Further, in Weight ~ Weeks × Diet + (1 + Weeks|Subjects), including Weeks as a random effect will then allow all subjects to have uniques slopes which can be seen as a step further than just including WeeksxDiet but no random effect for Weeks?
thank for great video. I have a question that my data is repeated measurment, but the output values are not linear with the predictor, which method should I use?
If your problem fits the example in this video, except that there seems to be some nonlinear trend in the data, then you can use a nonlinear mixed effect model. I aim to create such a video next year.
Thanks for the great video! I'm trying to code along in R, but I can't replicate the simple regression. Is your code available anywhere? Or could you show me the data you used to fit the simple regression? That would be very helpful!
At 2:25, you see how the data should be set up.
Then you just run:
lm(Weight~Weeks,data=df)
or:
summary(lm(Weight~Weeks,data=df))
Thank you!
What happens when you have multiple fixed effects for example weeks and exercise? Having difficulty visualizing that.
In the example, I have weeks and diet. Do you mean what would happen if you instead have two continuous variables? If so, you need a three dimensional plot, where you instead fit a plane to the data.
@@tilestats Thank you. yes I am asking about two continuous variables. I'm aware that they would form a plane but could this still be visualized in a similar way as above in two dimensions?
I have seen examples like this, m
Why using the lme4 package instead of nlme with lme function? Really like you videos!! Thanks so much
nlme also works fine. I have used both but I think the way you enter your model is a bit simpler in lme4.
I really appreciate your clear explanation on the concept of linear mixed effects models. Could you or someone clarify which data format should be set for using lmer? I think 'long' format according to "weeks" would be appropriate. Right?
Set up the data frame as shown at 2:27
Thanks@@tilestats
I am migrating from SAS and mostly struggling with code. How would I include a second or third random slope in R?
What if I want to do a pre- & post- tests for the weight? (two measurements). SPSS can only input one dependent variable in the linear mixed model.
Can't you do as in the example but where you only have two time points (pre and post)?
At 2:48, why I only got one intercept instead of multiple?
6:57 I don't understand how we can now assume that the Groups have one random intercept. That would mean each group has their own distribution, which is not the case because as you said all 4 subjects are randomly sampled out of one distribution.
When we include Diet in the model, we can test if the two groups have different intercepts. Thus, we then no longer assume that all individuals are sampled from the same distribution.
@@tilestats thanks! I just realised the difference between groups and clusters. Now everything is clear!
If I have different Genes count in place of weight How can I used that for 10-20 types Genes. Should I use a loop to get results for each Genes count changes over the week by the influence of Diet or is that any possibility to create model for all Genes at a time. One more question you have not explained how we will know how much changes happen in weight over the weeks while comparing both the diet
I would suggest that you look into how to set up a repeated measure design in packages, like DESeq2. limma or edeR, that have been developed to deal with gene count data.
@@tilestats no actually I want to see How Gene counts have changed over the weeks by the effects of the diet
Since you have counts, you should use GLMM instead of LMM.
@@tilestats ok sir
Many thanks for your excellent videos. But the lecture over it is not much understandable. So, as feedback, can you possibly give your lecture's script to a text-to-speech engine (bot) to read it aloud over the video? I am saying this because you pronounce many words very badly, severely ruining your awesome video and badly distracting the audience. The worst of them is the word "model" which is pronounced by you as "mole" or "mowel". There are many such odd pronounciations. I would be very grateful if you could give the text to a bot to speak it. Thanks again for your awesome videoes.
What a cringy comment dude. Thats his pronunciation, which is fine and understandable, what do you mean "you get distracted" by it? Pay your respect to the man and stop distracting us with your useless and offensive comments.
Hello good, what is your email? I would like to ask you about an exercise that I have